【作者】 牛路辰；

【导师】 谢平；

【作者基本信息】 西南财经大学 ， 金融学， 2013， 博士

【摘要】 近年来,重复博弈框架下的声誉理论不断丰富,在不同领域有了比较深入的应用,在产业经济学中Kreps对重复进入阻挠博弈的研究已经成为基础性文献,在宏观货币政策可信性上Vickers对通胀-预期博弈的研究也提供了一条新路径,然而对于借贷交易市场中声誉的专项研究却不多见,也不系统。实际上,借贷市场中声誉的作用更为显著和重要。传统货币银行学是建立在信用这一核心概念之上的,然而“信用”是作为先验概念引入,并未加以数量化、精确化。关于借贷声誉的研究有可能为货币银行理论提供更为坚实的基础,简单说就是提供一种“可计算信用”。本文大致分为三个部分：第一部分为理论篇,包括第2至第6章,构建介绍借贷声誉的基础理论；第二部分为案例篇,包括第7至第9章,以人民银行征信中心为例对借贷声誉进一步作深入研究；第三部分为结论与展望,包括第10章,给出基本结论并指出理性选择范式下借贷声誉研究的不足和未来的方向。基于这一安排,本文首先对借贷市场上关于声誉的理论进行评述,随后在重复博弈的框架下分两种情况,分别采用不完美公共监督模型和完美公共监督声誉模型对信贷声誉均衡及其性质进行讨论。第一种情况是讨论借款人行动与信号空间不完全同构时,借贷声誉均衡的特征。为使讨论有意义,假设Rf≤r<Rs,即借贷资金利率高于项目失败时的收入,但是低于项目成功时收入(存在Rs≤r的情况,我认为这属于商业银行不能识别有效需求,这种情况下声誉机制同样起作用)。单次借贷博弈中,借款人类型的信息不对称,贷款人对借款人采取的行动无法观测,受噪声干扰的项目收益随机分布,借贷交易中会出现逆向选择和道德风险,以及对状态的有成本证实。如果借款人行为在统计上可根据公共信号加以识别(这一假定与事实相符),在重复借贷博弈中,借款入不仅考虑本次交易的收益,还要考虑本次交易对于未来交易的影响,在违约获得的当期收益低于因偿还而获得的未来收益流的贴现时,借款人会建立声誉。然后在博弈接近结束时利用声誉获利。在声誉具备价值的条件(当期收益低于未来收益流贴现和)成立时,声誉对于缓解借贷交易的逆选择、道德风险具有一定作用。如详加考察,则会发现借贷声誉的建立是一个效应不断加强的正反馈过程,借款人较高努力程度降低项目风险,因而获得较低利率,声誉价值提高,借款人更加重视自己的声誉,更加努力工作,利率进一步降低。如果把声誉看做一种专用性资产,在借款人的生命周期中声誉的价值并非固定不变的,它随外界诸多因素的影响而变化。戴蒙德建立了声誉机制的动态模型,刻画了声誉价值的变动。借贷声誉对于借贷资金价格(市场利率)和借贷交易数量的会产生影响。单期交易中,承诺型借款人总是选择安全项目,策略型和普通型借款人的最优选择为风险项目。而重复交易过程中,通过声誉机制一方面不断地剔除掉选择风险项目的借款人中的失败者,另一方面声誉具备价值的情况下,策略型借款人也会选择安全项目,声誉这一隐性激励机制,起到了消弱借款人产生道德风险的动机的作用。以上两方面作用使风险项目在借贷市场上的比率下降,从而降低银行业金融机构的不良贷款率。假设贷款人之间存在竞争,只能获得行业平均的利润率,则借款人支付的贷款利率会缓慢下降。我构造了一个模型,解释声誉可以通过影响借款人申请贷款的概率缓解逆向选择的程度。由于配给是借贷市场的主要特征,声誉通过对逆向选择、道德风险、监督成本,进而影响配给的程度,最终对借贷交易价格和数量产生作用。基于信贷配给程度缓解,商业银行对每一借款人放款数量会增加。另外,如果借款人群体的整体风险状况下降,这会促使资本金一定的商业银行扩大贷款规模。第二种情况是讨论借款人行动与信号空间同构时,借贷声誉均衡的特征。为简化讨论,假设r≤Rf,即借贷资金利率低于项目最低收入。对借贷重复博弈的模型化处理中,存在两种类型的借款人,承诺类型和策略类型,虽然项目收入足以偿还贷款本息,如果外在强制执行机制无效,策略类型借款人仍有可能违约(策略型参与人的机会主义行为)。假定强制执行机制无效,单期博弈下策略型借款人的最优选择是违约。在重复博弈并且声誉具备价值的情况下,策略型借款人会选择偿还贷款,以获得良好声誉,只有在博弈接近结束时期,策略型借款人则又会重新选择违约。这是对第一种情况的补充,是特殊情况下借贷声誉发生作用的一种解释。在契约理论中,声誉是一种自执行机制,声誉机制和外在第三方强制执行机制之间存在复杂的替代与互补关系,通过以上模型我们给出了清晰的理论解释。实际上,作为隐性激励,借贷声誉可以降低监督成本和证实成本。在法律制度不健全,契约执行易受外在因素干扰和影响的发展中国家,声誉的这一作用尤其重要。其次,给出借贷声誉机制的一个典型例子——征信中心。人民银行征信中心负责建设和运营金融信用信息基础数据库。截至2012年12月31日,个人信用信息基础数据库共采集的自然人数达到82258.4万人,其中有信贷记录的人数28934.7万人,全年累计查询27427.2万次,日均查询74.9万次。企业征信系统共收录企业与其他组织个数为1858.8万个,其中有信贷记录的企业和其他组织388.1万个,全年累计查询9733.1万次,日均查询26.6万次。这一世界上最大的征信数据库在经济金融活动中起到了何种作用?又是如何发挥作用的?其效率受哪些因素影响?需要从哪些方面加以改进?本文从声誉的视角,采用有信息成本的借贷声誉模型,在理论上解释征信活动对信贷交易产生作用的机制及其效果。征信活动中,信用信息的采集、加工、匹配和对外提供查询都需要成本。不同的市场竞争结构和制度框架下贷款人或征信机构处理信息会产生不同的成本结构。征信机构在信息处理上具有比较优势。征信机构对外提供信息的数量和收费方式也有不同方式,因而对借款人和授信机构的行为都会产生不同影响。我通过对不同种类的信息搜集成本函数进行对比,分析不同种类成本对借贷声誉均衡的影响。在对征信作用机制分析的基础上,研究征信规制对于征信活动的影响,着重研究征信报告应用范围,强制报送数据,数据保存期限等因素对于征信作用的影响。再次,得出研究的结论,主要有：借贷声誉是一种专用性资产,能够改善借贷市场绩效,降低借贷资金价格,扩大借贷交易数量；外在执行机制无效时借贷声誉在一定程度上促进了契约执行；征信机构在信息处理上具有优势,能够降低信息搜集成本,改善市场均衡；征信规制能够通过借贷声誉均衡的改变影响借贷市场；借贷声誉发挥作用需要一系列条件。同时,本文指出了理性选择范式的不足,介绍心理学、社会学等对于信用、声誉的研究。在展望一节中,认为在理性不足和纯粹不确定性情况下借贷声誉更加值得研究；在大数据时代,借贷声誉所发生得深刻变化值得关注。本文对于借贷声誉研究的不足有：(1)借贷声誉理论的内容并未完全概括,模型细节处理不尽完美,尤其是不完美公共监督模型还有需要完善的地方,对于博弈论工具掌握的不是很熟练；(2)理性选择范式造成的理论对现实的解释力不足,解释中没有人类行为心理、不确定性的内容,而这是现代金融理论中最常提起的；(3)对于互联网大数据时代征信活动的新发展把握不够。本文的创新之处：(1)根据不完美公共监督的借贷声誉模型,对比借贷市场的配给均衡与声誉均衡,解释了声誉对借贷市场绩效的改善,指出借贷声誉价值的来源、动态特征和发生作用的条件；(2)构造了模型,解释声誉可以通过影响借款人申请贷款的概率以缓解逆向选择的程度,进而影响贷款人对每一借款人放款数量,借贷交易量会增加；(3)构造完美公共监督的借贷声誉模型,解释外在执行机制无效时借贷声誉发生作用的机制和条件；(4)将有信息搜集成本的声誉模型拓展至不完美公共监督的情形,应用于借贷交易中,将征信处理为不同的信息搜集成本函数,研究对借贷市场绩效的影响。更多还原

【Abstract】 In recent years, theory of repeated game and reputation prospered and became in-depth application in different fields. Krep’s work of "entry-deterrence" has become the fundamental literatures in industry organization economics, Vicker’s study of "inflation-expectation" also provides a approach to credibility of monetary policy. But literatures about credit reputation is rare, also not system. In real world, credit reputation play a significant and important role in credit market, researchs on credit reputation is likely to provide a more solid foundation for monetary banking theory, that is to provide a "calculable credit". The core concepts of "credit" that traditional monetary theory is based on is a priori, not an accurate one.My article roughly divided into three parts:part Ⅰ is the theory, including chapter2to chapter6, build reputation models to introduce basic theory; part Ⅱ is case study, including chapter7to chapter9, with the credit referring center of PBC as an example for further study; part Ⅲ is the conclusion and prospect, including chapter10, gives the basic conclusion and points out the shortage of my research and the direction of the future.For this purpose, in part Ⅰ I summarizes theory of reputation on the credit market, then under the framework of repeated game, I divide it into two circumstance, imperfect public monitoring model and perfect public monitoring model, discuss its properties.The first circumstance is to discuss credit reputation equilibrium when the borrower’s action are not observed. To make discussion meaningful, let Rf<R≤RS. When game play only once, because information on borrower’s type is asymmetric, the borrower’s actions cannot be observed, project returns is disturbed by the noise and random distributed, there are adverse selection and moral hazard in credit transactions. Assumes that the borrower behavior verified statistically according to public signal, in repeated game, borrowers not only consider the current payoff, but also consider the impact of his action on future transactions, conditional on current payoff when default is less than discounted expected revenue stream, the borrowers will build a reputation. Only at the end of the game borrower exploit reputation. when reputation has value, it can alleviate adverse selection and moral hazard. Credit reputation building is a positive feedback process, the borrower in higher degree of efforts make risk lower, and get lower interest rates, thus improve reputation value, borrowers pay more attention to his reputation, work harder, and interest rates go down. As a specific assets, the value of reputation in the borrower’s life cycle is not fixed, it varies with many factors. Diamond has established a model, depict the reputation value’s dynamic. In chapter3, the main result is to explan the mechanism, conditions, and length of final period.Credit reputation influce interest rates and transaction volume. When game play once, borrower of commitment type always choose safe project, strategy type and general type choice risk one. When game play repeated, borrower choose risk project get failure and out of the original pool of borrower, if reputation has value, strategy type would choose safe project, through the reputation mechanism, moral hazard incentives weakened. From the two reason above, ratio of default in credit markets decline. Under the condition of competition,lenders can only get the average profit, the interest rate borrower pays will be slow down. I constructed a model to explain influence of reputation to the probability of a borrower apply for a loan. As a result, loan volume of commercial banks to a borrow will increase.The second circumstance is when the borrower action and the signal space is isomorphism. To simplify the discussion, let R≤Rf,the income of the project always afford the interest. In the model, there are two types of borrowers, commitment type and strategy type, although the project income is sufficient to repay the loan, but the strategy type may default (opportunism behaviors). Assume that external enforcement mechanism is invalid, the optimal strategy of strategy type is default. In repeated game, when reputation have value, strategy type choose to repay the loan, in order to build good reputation, and only in the final period of the game, strategy type borrowers would default. This is supplement to the first case. In contract theory, the reputation is a self-enforcement mechanism, relationship of reputation and third-party enforcement is alternative and complementary. In developing countries, the court system is not robust, contract execution is vulnerable to external interference, this effect of reputation is particularly important.Part Ⅱ, we study a case of reputation mechanism, Credit Referring Center of PBC. Credit Referring Center of PBC is responsible for the construction and operation of nation financial credit information basic database. As of December31,2012, the individual information basic database collect information of822.584million people, in the database289.347million have credit records, annual accumulative queries is274.272million, average daily queries749000times. Enterprise information basic database collect for18.588million, including3.881million have credit records, annual accumulative queries is97.331million, average daily queries is266000times. As the largest credit reporting database of the word, what role had Credit Referring Center of PBC played in economic and financial activities? And how?What factors influence its efficiency?how to improve its efficiency?I use a reputation model with information cost, theoretically explain the activity of Credit Bureau. Credit reporting activities, credit information collection, processing, matching, and query all have costs, with different market competition structures and regulation framework, information processing will have different cost structures, charge policy also has difference, borrower’s behavior may be changed. Through the different approach of information gathering, I compared the cost function and the influence to reputation equilibrium. Use this I study the effect of credit reporting regulation on credit activities, mainly credit report application scope, forced to submit the data, data expiration.Part Ⅲ conclusioned, mainly include:borrowing reputation is a kind of specifica assets, it can improve market performance, lower the cost of capital, expand transaction volume; Reputation as a self-enforcement mechanism is useful when external enforcement mechanism is invalid; Credit bureaus have advantages in information processing, can reduce the costs of information gathering, improve market performance; Credit reporting regulation can change reputation equilibrium affect credit market, credit reputation play a role requires a series of conditions. At the same time, I points out the shortages of my rational choice paradigm, introduction similar study of psychology, sociology, etc. The outlook section, I think in bounded rationality and pure uncertainty, credit reputation is worthy of study; What happened in the era of big data, credit reputation have profound changed. Deficiencies of the study are:(1) credit reputation theory has not fully sum up, detail of the model is not perfect;(2) the explanatory power of rational choice paradigm is limited, no human behavior psychology, uncertainty in the explanation, but as is the most commonly mentioned in modern financial theory;(3) game theory is not very skilled.The innovation of my paper:(1) built the imperfect public monitoring credit reputation model, and contrast the performace of credit market, and explains the improvement of credit market under reputation equilibrium, pointed out the source of credit reputation value, dynamic characteristics;(2) constructed model, explain that reputation can affect the probability of application to alleviate the degree of adverse selection, and then loan volumes will increase;(3) built perfect public monitoring reputation model to explain the effect of reputation mechanism when external enforcement mechanism is invalid;(4) expand model to imperfect public monitoring, applied to the credit market, for different information collecting cost function, study its influence on credit market performance.更多还原